/**
* Initializes the scanner.
*
* @param f input file
* @param pr database properties
* @param frag allow parsing of document fragment
* @throws IOException I/O exception
*/
XMLScanner(final IO f, final Prop pr, final boolean frag) throws IOException {
input = new XMLInput(f);
fragment = frag;
try {
for (int e = 0; e < ENTITIES.length; e += 2) {
ents.add(token(ENTITIES[e]), token(ENTITIES[e + 1]));
}
dtd = pr.is(Prop.DTD);
chop = pr.is(Prop.CHOP);
String enc = null;
// process document declaration...
if (consume(DOCDECL)) {
if (s()) {
if (!version()) error(DECLSTART);
boolean s = s();
enc = encoding();
if (enc != null) {
if (!s) error(WSERROR);
s = s();
}
if (sddecl() != null && !s) error(WSERROR);
s();
final int ch = nextChar();
if (ch != '?' || nextChar() != '>') error(DECLWRONG);
} else {
prev(5);
}
}
encoding = enc == null ? UTF8 : enc;
if (!fragment) {
final int n = consume();
if (!s(n)) {
if (n != '<') error(BEFOREROOT);
prev(1);
}
}
} catch (final IOException ex) {
input.close();
throw ex;
}
}
/**
* Scans a markup declaration. [29]
*
* @return true if a declaration was found
* @throws IOException I/O exception
*/
private boolean markupDecl() throws IOException {
if (consume(ENT)) { // [70]
checkS();
if (consume('%')) { // [72] PEDecl
checkS();
final byte[] key = name(true);
checkS();
byte[] val = entityValue(true); // [74]
if (val == null) {
val = externalID(true, false);
if (val == null) error(INVEND);
}
s();
pents.add(key, val);
} else { // [71] GEDecl
final byte[] key = name(true);
checkS();
byte[] val = entityValue(false); // [73] EntityDef
if (val == null) {
val = externalID(true, false);
if (val == null) error(INVEND);
if (s()) {
check(ND);
checkS();
name(true);
}
}
s();
ents.add(key, val);
}
check('>');
pe = true;
} else if (consume(ELEM)) { // [45]
checkS();
name(true);
checkS();
pe = true;
if (!consume(EMP) && !consume(ANY)) { // [46]
if (consume('(')) {
s();
if (consume(PC)) { // [51]
s();
boolean alt = false;
while (consume('|')) {
s();
name(true);
s();
alt = true;
}
check(')');
if (!consume('*') && alt) error(INVEND);
} else {
cp();
s();
// check(')'); // to be fixed...
while (!consume(')')) consume();
// input.prev(1);
occ();
}
} else {
error(INVEND);
}
}
s();
check('>');
} else if (consume(ATTL)) { // [52]
pe = true;
checkS();
name(true);
s();
while (name(false) != null) { // [53]
checkS();
if (!consume(CD)
&& !consume(IDRS)
&& !consume(IDR)
&& !consume(ID)
&& !consume(ENTS)
&& !consume(ENT1)
&& !consume(NMTS)
&& !consume(NMT)) { // [56]
if (consume(NOT)) { // [57,58]
checkS();
check('(');
s();
name(true);
s();
while (consume('|')) {
s();
name(true);
s();
}
check(')');
} else { // [59]
check('(');
s();
nmtoken();
s();
while (consume('|')) {
s();
nmtoken();
s();
}
check(')');
}
}
// [54]
pe = true;
checkS();
if (!consume(REQ) && !consume(IMP)) { // [60]
if (consume(FIX)) checkS();
quote = qu();
attValue(consume());
}
s();
}
check('>');
} else if (consume(NOTA)) { // [82]
checkS();
name(true);
s();
externalID(false, false);
s();
check('>');
} else if (consume(COMS)) {
comment();
} else if (consume(XML)) {
pi();
} else {
return false;
}
s();
pe = false;
return true;
}